Merge pull request #788 from PrefectHQ/add-ex

Add example of dynamic mapping w/ retries

examples/retries_with_mapping.py

@@ -0,0 +1,35 @@
+"""
+This example showcases a few basic Prefect concepts:
+    - the ability to "map" tasks across the dynamic output of upstream tasks; additionally, it is not required that the
+        upstream task actually passes data to the downstream task, as this example demonstrates
+    - task retries: `flow.run()` will perform retries, on schedule, for all tasks which require it,
+        including individual mapped tasks
+
+This flow first generates a list of random length, and them maps over that list to spawn a dynamic number of
+downstream tasks which randomly fail.  The takeaway here is that we don't have to know a-priori how many mapped tasks
+will be created prior to execution!  Additionally, each failed mapped task will retry on its own.
+"""
+import random
+from datetime import timedelta
+
+from prefect import task, Flow
+
+
+@task
+def generate_random_list():
+    n = random.randint(15, 25)
+    return list(range(n))
+
+
+@task(max_retries=3, retry_delay=timedelta(seconds=0))
+def randomly_fail():
+    x = random.random()
+    if x > 0.7:
+        raise ValueError("x is too large")
+
+
+with Flow("random-mapping") as f:
+    final = randomly_fail.map(upstream_tasks=[generate_random_list])
+
+# should see logs suggesting that some tasks are failing and retrying
+f.run()